MedicalDataAI / demoshow

AI code demo

ThyroidUS AI

AI code demo, paper name Lymph node metastasis prediction from primary breast cancer ultrasound images using deep learning

If you use this code in your research, consider citing:

@article{
  title={xxxx},
  author={xxxx},
  journal={xxxx},
  year={xxxx},
  publisher={xxxx}
}

Prerequisites

• Ubuntu 18.04 with Nivida 2080Ti
• Python 3.6 with dependencies listed in the requirements.txt file

sudo pip install -r requirements.txt

Running

1. clone the repo to local directory

   git clone https://github.com/MedicalDataAI/demoshow.git

2. download the weight file of the trained model into the folder of "./models"

   wget 'https://drive.google.com/file/d/1KWAmU7hbp-byJ9tM8Mh6_I9_85mOTYsD/view?usp=sharing'
   wget 'https://drive.google.com/file/d/1KWAmU7hbp-byJ9tM8Mh6_I9_85mOTYsD/view?usp=sharing'
   wget 'https://drive.google.com/file/d/1KWAmU7hbp-byJ9tM8Mh6_I9_85mOTYsD/view?usp=sharing'
   wget 'https://drive.google.com/file/d/1KWAmU7hbp-byJ9tM8Mh6_I9_85mOTYsD/view?usp=sharing'

4. use the trained model to predict the data of Clinical, BMUS, CDFI (NOTE: to replace the parameter of path with proper location of model file)

• Predict clinical data (NOTE: to replace the data in "./data/clinical.csv") into "./res/result_clinical.csv"

   python3 clinical_lr.py

• Predict BMUS data (NOTE: to replace the images in "./data/bmus/*") into "./res/result_bmus.csv"

   python3 bmus_cnn.py

• Predict CDFI data (NOTE: to replace the images in "./data/cdfi/*") into "./res/result_cdfi.csv"

   python3 cdfi_cnn.py

• Predict ensemble data (Prompt for input the risk of Clinical, BMUS and CDFI from clinical_lr.py, bmus_cnn.py and cdfi_cnn.py.)

   python3 ensemble_bagging.py